Sie sind auf Seite 1von 25

CBMH 13.

2_2nd 3/10/03 5:53 pm Page 81

Criminal Behaviour and Mental Health, 13, 81–105 2003 © Whurr Publishers Ltd 81

A developmental test of the general


deviance syndrome with
adjudicated girls and boys using
hierarchical confirmatory factor
analysis
MARC LE BLANC 1 AND CHRISTIANE BOUTHILLIER 2 1 École de
psychoéducation and École de criminologie, Université de Montréal,
Québec, Canada; 2École de psychoéducation, Université de Montréal

ABSTRACT
Background Over the last 40 years, numerous studies have proposed that various
deviant behaviours are part of a latent construct now labelled ‘general deviance’ by
criminologists or ‘problem behaviour’ by psychologists. During that period, many
studies have documented the presence of specific forms of deviance. However, no
study has tested these two opposing views simultaneously, particularly with longitu-
dinal data.
Aims The objectives of this paper are the cross-cultural replication of the construct of
general deviance for a French-speaking adjudicated sample of girls and boys and,
specifically, the developmental replication of the general deviance syndrome.
Method The age of onset is used as a developmental indicator of deviance instead of
measures of participation or frequency.
Results The results of EQS hierarchical confirmatory factor analyses supported the
existence of the construct of general deviance. In addition, there is no gender gap in
the structure of the general deviance syndrome. This paper reports a comprehensive
test of the general deviance syndrome because of the use of 45 deviant behaviours and
nine types of deviance classified into four categories.

Introduction

The criminological literature has swung from a unidimensional to a multidi-


mensional approach to the measurement of self-reported deviance. Many
researchers have used a scale of general deviance and theoreticians have
supported that practice with general theories. Other criminologists have studied
specific forms of deviant behaviour. This paper proposes an enlarged model of
deviant behaviour that is simultaneously unidimensional and multidimen-
sional. Our demonstration uses a developmental descriptor of self-reported
CBMH 13.2_2nd 3/10/03 5:53 pm Page 82

82 Le Blanc, Bouthillier

deviance, the timing of onset, rather than participation or frequency. In


addition, we test the possibility that there is a gender gap in the structure of
general deviance.

Deviance, whole or parts

Early on, Nye and Short (1957) proposed an omnibus scale of self-reported
delinquency. This scale was misnamed because their 21-item questionnaire and
their seven-item scale were composed of a large array of behaviour. The
categories included theft, vandalism, sex, alcohol and drug use, fighting,
aggression, joyriding, dangerous driving, and misbehaviour at school and in the
family. At the same time, many criminologists continued to misname their
measure of deviance as delinquency. Their scales involved more serious delin-
quent acts but, in many cases, they were still composed of many types of
delinquency, substance use and other deviant behaviour. Elliott et al. (1985)
avoided that pitfall and theirs is probably the measure most often used in
criminological research. Their 24-item scale of general delinquency has only
four questions that are not clearly criminal offences or defined as such in
criminal codes: runaway, joyriding, gang fights and sexual intercourse. Most
criminologists followed the omnibus road indicated by Nye and Short. General
theorists, such as Gottfredson and Hirschi (1990) supported that practice; they
argued that all forms of deviant behaviour were analogous phenomena.
Empirical research identified a latent construct as illustrated by the studies
outlined in Table 1. Using various statistical methods, they showed that a large
variety of deviant behaviours represented the notion of deviance. The
construct was supported for officially recorded crimes (Parker and McDowell,
1986). In addition, numerous empirical studies of self-reported deviance
concluded similarly since the landmark study by Jessor and Jessor (1977).
Replications have taken place with children (Cappaldi and Patterson, 1989;
Gilmore et al., 1991; Cheong and Raudenbush, 2000; Farrell et al., 2000), with
samples of adolescents (Johnston et al., 1978; Donovan and Jessor, 1985;
Brownfield and Sorenson, 1987; Donovan et al., 1988; McGee and Newcomb,
1992; Emler and Reicher, 1995; Benda and Corwyn, 1998; Bartush et al., 1997;
Ary et al., 1999; Corwyn and Benda, 1999) and with young adults (Donovan
and Jessor, 1985; Osgood et al., 1988; McGee and Newcomb, 1992; Ullman
and Newcomb, 1999). Only Dembo et al. (1992) and Le Blanc and Girard
(1997) replicated the general deviance construct with samples of high-risk
adolescents.
Some of these studies used national samples of US residents, while most
of the other samples were local (counties, cities or regions) and only two
studies sampled adjudicated adolescents. However, three replications did
take place outside the Unites States: England (Emler and Reicher, 1995),
New Zealand (Bartush et al., 1997), and Montreal (Le Blanc and Girard,
1997), but none was conducted with specific ethnic groups even if some
CBMH 13.2_2nd 3/10/03 5:53 pm Page 83

Development test of the general deviance syndrome 83

samples were predominantly composed of African-Americans (Farrell et al.,


2000). Only one study tested the deviance syndrome for age groups during
adolescence and adulthood (McGee and Newcomb, 1992; Ullman and
Newcomb, 1999). Even if samples were recruited in the 1970s, 1980s and
1990s, only one study compared directly equivalent samples from two
decades and concluded that there was a unique latent factor for the same set
of behaviours 20 years apart (Le Blanc and Girard, 1997).
Most of the replications had in common the use of some forms of grouping
analysis. Ordinary factor analysis, confirmatory factor analysis and structural
equation modelling were the most often used statistical techniques. The
grouping procedure is applied on a certain number of measures representing
various forms of deviance, each of them composed of different items in
different numbers. Only six studies used specific behaviours. Twelve forms of
deviant behaviour were considered: theft, vandalism, physical and non-
physical interpersonal violence, sexual violence, school problems, parental
defiance, sexual intercourse, forms of acting out, use and traffic of a large array
of drugs, use of alcohol, and use of tobacco. Two studies used nine of these
forms of deviance, six studies eight, three seven, two six, and six five or less. In
addition, all 21 studies employed some measures of theft and interpersonal
violence; most considered measures of drug use or traffic, 14 included
vandalism and alcohol use, and nine involved school problems. Sex and
parental defiance are present in eight studies. Sexual violence, dangerous
driving and use of tobacco are very rarely considered. Some studies included
measures that are not intrinsically deviant behaviour such as academic results,
going to church, GPA, social conformity and so on. The diversity of infor-
mants used in the statistical analysis is not a strong point of these replications.
Only one study employed multiple informants. Finally, 17 of the 21 studies
used mixed-gender samples. However, only three of them tested and confirmed
the general deviance model by gender (Jessor and Jessor, 1977; Donovan and
Jessor, 1985; Donavon et al., 1988).
In sum, a general deviance construct was always identified by these studies.
This result was independent of the informant, the set of deviant behaviours,
the statistical method, the nature of the sample, the historical period and the
site of the study. The questions that are still to be resolved are, first, what
should be the behavioural content of that general deviance construct because
existing studies use various sets of deviant behaviours and, second, is the
general deviance syndrome replicable with developmental descriptors of
deviance?
Very early on, the Nye and Short (1957) unidimensional scale was criti-
cized on the ground that delinquent behaviour was multidimensional.
Researchers adopted two strategies. First, some researchers, for example Scott
(1959), developed ad hoc scales representing different forms of deviant
behaviour. This type of scale is still used in the contemporary literature.
Second, with the advent of computers, some researchers started to use cluster
84
CBMH 13.2_crc

Table 1: Summary of studies on the latent construct of deviant behaviour

Studies Samples Scales and behaviours considered Statistical techniques


6/10/03

Ary et al. 608 14- to 17-year-old adolescents (35% Antisocial behaviour: vandalism (3), threats Structural equation
(1999) males, 65% females from large metropolitan (2), theft (3) school (2); risky sexual behaviour modelling
area in Pacific Northwest (4), substance use (3), academic failure (2)

Bartusch et al. New Zealand birth cohort; 976 males (52%) Fights, destroys property, disobedient, Second-order
(1997) and females (48%) assessed at 5, 7, 9,11, 13, bullies, irritable, temper tantrums confirmatory factor
1:18 pm

Le Blanc, Bouthillier

and 15; teacher, parent, and self-reports models

Benda & Corwyn 1093 youths (46% males and 54% females) Number of sexual partners, marijuana use and Logistic and OLS
(1998) aged between 13 and 20 years, recruited from heavy alcohol use. Scales of property crimes, regression analysis
high schools of Arkansas and Maryland crimes against persons, status offences, and property,
Page 84

persons and status offences

Brownfield & Sorenson 2174 males, aged between 15 and 18, from the Six self-report delinquent behaviours Latent structure
(1987) Seattle area in 1978–79 analysis
(Hindelang et al., 1981).

Cappaldi & Patterson 102 fourth-grade boys recruited in 1983–83 in Scales of coercive, overt and covert child Correlation, scale
(1989) three major school districts of Eugene-Springfield. behaviours. analysis and factor
Replication with 104 fourth-grade boys recruited analysis
in 1984–85 in same locations. Child self-reports
and parents’ and teachers’ reports

Cheong & Raudenbush 2177 children aged 9–15 in 79 urban CBCL/4-18: 20 items of the Aggressive Item response model
(2000) neighbourhoods in Chicago between 1994 and Behaviour Scale and 13 items of within a hierarchical
1997; nearly half were girls, 43% Latino and 37% delinquent behaviour scale model
African-American, various SES

Corwyn & Benda 1626 public school students (Black 61%, Daily tobacco use, last six months alcohol and Multiple contingency
(1999) females 53%) aged between 12 and 18 from a marijuana use, age of first sexual intercourse table analyses
southern metropolitan area

Dembo et al. 201 youths (148 males and 53 females aged 15 Five scales of self-reported delinquent Structural equation
(1992) years old) detained in Tampa between December behaviours, a scale of alcohol use, marijuana and modelling
1986 and April 1987. Reinterviewed twice at hashish use, and urine analysis for cannabinoids.
4-month periods

Donovan & Jessor First study: 102 males and 142 females in 7th, 8th A scale of general deviant behaviour, an item on Factor analysis
(1985) or 9th grade in 1969 and 84 male and 100 female drunkenness, drug use, and sexual experiences
college students in 1970), in a small city in Colorado.
Analysis on two data waves
Second study: 1208 boys and 1444 girls, senior high A self-report scale of general deviant behaviour, Factor analysis
school students in 1978 from 48 states and the items on smoking, drunkenness, marijuana, and
district of Columbia other illicit drugs
CBMH 13.2_crc

Third study: 186 males and 242 females in middle A self report scale of general deviant behaviour, Factor analysis
and late twenties. Subjects of first study items on drunkenness, marijuana, and other
illicit drugs

Donovan et al. 388 (162 males and 226 females) students in grades 11 A self-report scale of general deviant behaviour, an Factor analysis
6/10/03

(1988) and 12 in northeastern Colorado in 1985. item on drunkenness, marijuana, and sexual experiences

Farrell et al. Urban sample: 988 seventh graders in a public school Frequency of 26 behaviours: six drug use items, six Hierarchical confirmatory
(2000) system in a large city in the south-eastern US minor delinquency items, seven physical aggression factor analysis
(females 54%, Black 96%). Rural sample: 1895 sixth and seven non-physical aggression items
1:18 pm

graders from middle schools five counties in the southern


US (13% Black, 50% females).

Gillmore et al. 426 (52% males and 48% females) sixth-graders in 1986 A self-report index of school trouble, a teacher Confirmatory factor
(1991) from the Seattle school district report of aggressive behaviour, three self-report analysis
indicators of delinquency, and indicators of substance use
Page 85

Jessor & Jessor 432 (188 males and 244 females) 10th, 11th or 12th grade A self-report scale of general deviant behaviour, and Chi-square
(1977) high school students and 205 (92 males and 113 females) items on marijuana use, sexual intercourse’s, drinking,
junior college students in a small town in Colorado. problem drinking, and activism protest

Johnston et al. 1260 boys in the 10th grade from public high schools A self-report drug use index, two self-report Product-moment
(1978) across the United States. Analysis on 5 waves between delinquency indexes correlation
1966 and 1974

Le Blanc & Girard 438 adjudicated boys recruited in 1974 and 366 boys 20 self-reported deviant behaviours: vandalism, school Factor analysis
(1997) recruited in 1992 from the Montreal area aged 14 to 18 and family rebellion, minor thefts, major thefts,
aggressions, and drug use

McGee & Newcomb 7th to 9th graders recruited in 1976: 1634 first data wave, Self-reported index of drug use, social conformity, Hierarchical confirmatory
(1992) 654 late adolescence, 739 young adulthood, and academic orientation, sexual involvement, and factor analysis
614 adulthood criminal behaviour

Osgood et al. 717 young adults recruited in 10th grade in public high Self-reported measure of heavy alcohol use, indexes Structural equation
(1988) schools across the United States in 1966. Assessments at of criminal behaviour, marijuana use, other drug use, modelling
18, 19–20, and 21–22 and dangerous driving

Ullman & Newcomb 350 participants at ages 9, 13, 17, 21, and 35: 73% Four identical scales at each age: licit drug use Confirmatory factor
(1999) females and 33% ethnic and racial groups (beer, wine, liquor), illicit drug use (eight drugs from analysis
cannabis to cocaine), criminality (15 delinquent and
aggressive behaviours), social conformity (four items
Development test of the general deviance syndrome

for law abidance, liberalism, and religiosity)


85
CBMH 13.2_2nd 3/10/03 5:53 pm Page 86

86 Le Blanc, Bouthillier

(Kulik et al., 1968) and factor analysis (Quay and Blumen, 1963), to derive
scales representing forms of deviance. This literature has been reviewed by
Hindelang et al. (1981) for self-reported delinquency and by Cohen (1986) for
official delinquency. Since, there has been no significant advance in the search
for the multidimensional content of deviance. Hindelang and colleagues’
(1981) conclusion remained sound. They concluded that distinctive sets of
behaviours exist because these subsets of items are relatively highly correlated
with each other and relatively weakly correlated across subsets. In addition,
they showed that the set of behaviours identified for boys could also be
pertinent for girls.
In psychology, there was a consensus between factor analytic studies to
identify internalizing and externalizing problem behaviours (Achenbach,
1985). Within the externalizing dimension, the studies distinguished between
concealing and confronting behaviours (Loeber and Schmaling, 1985; Frick et
al., 1993) and oppositional and conduct problems (Frick et al., 1991).
Whatever the distinctions that theoreticians and researchers would make,
there was a consensus that we should distinguish at least six categories of exter-
nalizing behaviour as exemplified by Loeber et al. in their (1998) book. These
forms of conduct problems were delinquency, substance use, sexual intercourse,
physical aggression, concealing activities and conduct problems.
We can conclude that researchers could probably attain an easy consensus
on the following scales of deviance (see for example the international self-
reported instrument in Junger-Tas et al., 1994): general delinquency, serious
delinquency, vandalism, one or more theft scales (minor and serious for
example), one or more interpersonal violence scales (for example fighting and
aggression), substance use, two scales of problem behaviour, at school and in
the family, and scales concerning sexual activities and driving behaviour. Most
often, the papers that tested one or more theories used the first two scales,
while results concerning violence, substance use and problem behaviour are
rarely published in the same paper or the same journal even if they are
obtained from the same sample. The theoretical base for the multidimensional
content of deviance was not as well argued compared with theories supporting
the general deviance position. In criminology, there was a tradition for the
study of forms of criminal behaviour and there were specific theories for forms
of deviance, such as Durkheim’s classic theory of suicide. Cullen (1985)
reviewed these theories and argued that most general theories should include
structuring variables that may be responsible for the choice of a particular form
of deviance by the individual.
All the theories and empirical studies that we reviewed consider what has
been called generic descriptors (Loeber and Le Blanc, 1990; Le Blanc and
Loeber, 1998). In fact, the self-reported measures of deviant behaviour were
participation or frequency, and sometimes variety. In addition, most empirical
research was based on cross-sectional data and did not used developmental
descriptors of offending, such as onset, duration, aggravation. Le Blanc and
CBMH 13.2_2nd 3/10/03 5:53 pm Page 87

Development test of the general deviance syndrome 87

Loeber (1998) and Le Blanc (2002) reviewed different ways to identify a devel-
opmental sequence. The timing of onset of deviant behaviours was the most
fundamental way to look at that phenomenon. Data from a variety of studies
indicated that the age of onset of deviant behaviour was predictive of the
frequency, variety, seriousness and duration of deviance. In addition, studies
documented the fact that there was a developmental sequence among a large
set of deviant behaviours and different forms of deviance, particularly in terms
of onset.
Is there a gender gap in the development of general deviance? Lanctôt and
Le Blanc (2002) reviewed that question and concluded that girls displayed
fewer tendencies than boys to get involved in delinquency and serious
deviance on a long-term basis. However, they reported that it was more
common for them to start their delinquent activities later than boys even if
they displayed some problem behaviour earlier than boys. In addition, they
more often had first contact with the justice system once reaching adulthood.
In sum, even if the timing of deviance was different, these authors concluded
that the developmental process did not seem different.

A model of adolescent deviance

Figure 1 proposes a classificatory model of adolescent deviance that tries to


reconcile the unidimensional and multidimensional perspectives. Figure 1
suggests that adolescent deviance is composed of four major forms of deviant
behaviours. Following Loeber and colleagues, we can state that the first
category of deviant behaviour is overt or confronting, the interpersonal
violence dimension. The second category of deviant behaviour is covert or
concealing; it is composed of various forms of property crimes. The third
category of deviant behaviour is concerned with authority conflict or
rebellion: stubborn, defiant, and avoidant behaviours at home and in school.
In addition to these forms of deviant behaviour, we propose reckless behav-
iours, which are not part of Loeber and colleagues’ typology. These behaviours
concern substance use, sexual activity, disorderly conduct and motor vehicle
use. The first three of these categories are standard forms of deviant behaviours
included in most self-reported instruments. The last category is not always
present, but the Osgood et al. (1988) study has proved that dangerous driving
is part of the general deviance syndrome. Based on our reading of the crimino-
logical literature on existing scales of deviance, Figure 1 specifies that the four
major forms of adolescent deviance are composed of 10 types of deviant
behaviour: family rebellion, school rebellion, sexual activity, substance use,
motor vehicle use, disorderly conduct, vandalism, interpersonal violence,
fraud, and theft. These 10 types are operationalized through 61 deviant behav-
iours, as can be seen in Figure 1.
The developmental perspective is introduced in this paper through the age
of onset of each form and type of deviant behaviour. To understand the nature
CBMH 13.2_2nd 3/10/03 5:53 pm Page 88

88 Le Blanc, Bouthillier

of the development of the general deviance latent construct, Le Blanc and


Janosz (1998) proposed the metaphor of a river. There is a general deviance
river composed of four forms of deviant behaviour, named overt, covert,
authority conflict, and risk behaviours. These forms of deviance are manifested
by a number of currents. These currents have their origin in the individual and
his environment. These types of deviance can remain dormant during the life
course, as is the case for the abstinent. Alternatively, they can manifest
themselves in different ways at different ages during adolescence and beyond.
The timing of onset of each behaviour and type of behaviour becomes the
starting point of the appearance of a new current in the river. To illustrate this
point of view, Table 2 reports the average age of onset for the behaviours and
the categories of deviant behaviour. This table is organized according to the
final empirical model that is composed of nine types and four forms of
deviance instead of the theoretical model proposed in Figure 1 with 10
categories of behaviour and 61 individual behaviours.
Whatever the point of view on the controversy between the unidimen-
sional and the multidimensional nature of deviance, the empirical solution has
to come from the simultaneous test of these perspectives. This complete test
was never done. On one hand, the unidimensionalists were using scales of
types of deviance to demonstrate the existence of the latent construct of
general deviance. On the other hand, the multidimensionalists were grouping
individual behaviours into subsets without interest in the latent construct of
general deviance. The complete test would require the confirmation of the
general deviance syndrome starting with a large array of behaviours rather than
scales, subsequently identifying the structure of the deviance syndrome and its
unidimensionality. The present paper does just that but with a developmental
descriptor of deviance, age of onset, rather than cross-sectional descriptors
such as participation, frequency or variety. In addition, we will verify whether
there is a difference in the structure of the general deviance syndrome
according to gender.

Samples and methods

The above theoretical model is tested with an adjudicated sample of girls and
boys for the following reasons. First, participation in deviant behaviour was
very much higher in such a sample as compared with what was observed in a
representative sample (for a detailed description of such data see Le Blanc and
Fréchette, 1989). Second, a reasonable rate of participation was also available
in an adjudicated sample for more serious deviant and delinquent acts that
were very rare in the population, for example prostitution and rape. Finally,
because of these characteristics of a purposive sample, a larger array of deviant
behaviours can be considered and, consequently, more comprehensive models
can be tested.
CBMH 13.2_2nd 3/10/03 5:54 pm Page 89

Development test of the general deviance syndrome 89

Figure 1. A comprehensive hierarchical model of adolescent general deviance

The sample was composed of 656 adjudicated adolescents from 12 to 18


years of age. They lived in the Montreal area and they were adjudicated, about
equally under the Canadian Young Offenders Act (criminal code and statutory
offences) or under the Quebec Youth Protection Act (particularly for problem
behaviours such as drug abuse, truancy, incorrigibility and so on). In conse-
quence, we will use the term ‘adjudicated’ to characterize their legal status.
90
CBMH 13.2_2nd

Table 2: Onset data for behaviours, factors and higher order factors
3/10/03

General Deviance 7.11 2.25 ##

Second-order factor, final model Recklessness Authority conflict Covert delinquent behaviours Overt delinquent behaviours
10.85 2.6 ## 7.81 2.58 ## 9.97 2.82 ## 9.27 2.8 ##

First-order factor, final model Sex & alcohol Drug use Motor vehicle In school At home Fraud Theft Vandalism Violence
5:54 pm

2.5.6 8.9.10.11 38.39.40.41 12.13.14.15.16 18.19.20.21.22 27.28 30.31.32.34.35.36.37 2.41.54 47.48.49.50.51.52.53.


Le Blanc, Bouthillier

55.57.58.59.60

mean sd n mean sd n mean sd n mean sd n mean sd n mean sd n mean sd n mean sd n mean sd n


10.96 2.6 ## 13.47 2.4 ## 13.79 2.18 ## 8.38 2.7 ## 9.81 3.11 ## 13.00 2.3 ## 10.25 3.00 ## 11.93 3.03 ## 9.64 2.94 ##
Page 90

Initial model composition Sexual activity Substance use Motor vehicle In school At home Fraud Minor theft Major theft Vandalism Physical Intimidation
Variable label mean sd n mean sd n mean sd n mean sd n mean sd n mean sd n mean sd n mean sd n mean sd n mean sd n mean sd n
(1) Homosexual relations 14.49 3.82 35
(2) Heterosexual relations 12.87 2.1 489
(3) Homosexual prostitution 15.24 2.98 21
(4) Heterosexual prostitution 15.52 2.18 61
(6) Drinking alcohol 12.29 2.84 480
(7) Getting drunk 13.26 2.54 452
(8) Glue inhalation 13.88 2.02 88
(9) Soft drugs 13.66 2.33 413
(10) Chemical drugs 14.53 1.9 254
(11) Hard drugs 14.85 1.9 256
(12) Dealing drugs 15.16 1.88 276
(39) Driving without a permit 14.28 2.05 421
(40) Joyriding a car 14.57 1.93 331
(41) Joyriding a motorcycle 14.63 2.05 155
(42) Stealing and selling a car 15.11 1.8 189
(13) Disturbing class 8.97 3.11 450
(14) Impolite to a teacher 11.28 3.21 444
(15) Cheating 12.39 2.45 256
(16) Skip a class 12.91 2.45 433
(17) Skip school 12.34 2.55 435
(18) Stealing from school 13.1 2.51 157
(19) Running away 13.42 2.33 277
(20) Loitering at night 12.67 2.6 397
(21) Destroy family property 12.29 3.69 155
(22) Stealing from home 12.24 3.09 235
(23) Refusing to obey parents 10.86 3.71 419
(24) Hitting a parent 14.52 2.22 113
CBMH 13.2_2nd

(25) Hitting a sibling 11.69 3.88 179


(26) Gambling 14.1 2.28 400
(27) Crank calls 12.13 3.19 315
(28) Using false ID 14.73 1.54 338
(29) Entering without paying 13.36 2.27 403
3/10/03

(30) Setting a false alarm 12.79 3.02 207


(31) Shoplifting 11.08 3.16 390
(32) Stealing for less than $10 11.43 3.27 303
(33) Stealing between $10 and $100 13.37 2.48 327
(34) Stealing a bicycle 14.56 1.75 225
(35) Stealing more than $100 14.21 2.18 317
5:54 pm

(36) Buying or selling stolen goods 14.21 2.17 383


(37) Breaking and entering 13.44 2.64 369
(38) Burglary 14.39 2.08 292
(43) Breaking someone else’s property 12.45 2.27 292
(44) Breaking school property 12.29 2.92 143
(45) Breaking school premises 12.93 2.41 211
Page 91

(46) Breaking car parts 13.53 2.54 244


(47) Arson 12.99 3.29 110
(48) Fighting 10.52 3.25 459
(49) Hitting someone messing with you 12.08 3.17 341
(50) Want to hit someone scuffling you 13.07 3.07 246
(51) Using a weapon while fighting 13.84 2.49 304
(52) Gang fight 13.94 1.84 334
(53) Carrying a weapon 13.74 2.23 406
(54) Throwing objects at someone 13.1 3.09 211
(55) Accusing another of starting a fight 13.06 3.31 202
(56) Hitting someone for nothing 14.05 2.66 130
(57) Forcing someone to have sex 11.58 7.24 40
(58) Threatening someone for something 14.02 2.5 210
(59) Threatening someone for a favour 13.92 2.95 176
(60) Using physical force to dominate 13.24 2.94 203
(61) Encouraging others to hurt someone 13.72 2.84 208
Development test of the general deviance syndrome
91
CBMH 13.2_2nd 3/10/03 5:54 pm Page 92

92 Le Blanc, Bouthillier

The adolescents in this sample were placed in institutions of various security


levels or on probation because of a criminal act or supervised by a social worker
because of serious problem behaviour. They represented virtually the entire
population of adolescents who had been adjudicated in 1992-93. These adoles-
cents were interviewed twice, during the first and second halves of
adolescence, approximately at an interval of two years. The sample was
composed of 506 boys and 150 girls.
Of the 506 boys, 57% were adjudicated for a criminal offence and 43% were
wards of the court for severe problem behaviours. Before being adjudicated,
most of these boys were diverted by police or other agencies. They may also
have accepted voluntary measures under the Youth Protection Act or
diversion measures under the Young Offenders Act. Their average age was
15.87 years (3.6% 12 and 13 years, 14.6% 14 years, 20.4% 15 years, 25.3% 16
years, 25.1% 17 years, and 11.0% 18 years). Some who committed a crime
when they were 17 were adjudicated after their 18th birthdays. In sum, 88% of
the boys were 14, 15, 16 and 17 years old. These boys originated from broken
homes (only 16% lived with their two biological parents) and most of their
parents were born in Quebec (81%). Of the immigrant boys, 15% originated
from African and Arab countries, 12% were Haitians, 5% were Latin
Americans and 5% were from the Caribbean. Among the adjudicated boys,
68% were gang members, but only 22% were members of a structured gang.
Of the 150 girls, 17% were adjudicated for a criminal offence under the
Young Offenders Act and 83% were wards of the court for severe problem
behaviours under the Youth Protection Act. Before being adjudicated, most of
these girls were diverted by police or other agencies. They may also have
accepted voluntary measures under the Youth Protection Act or diversion
measures under the Young Offenders Act. Their average age was 15 years
(2.6% 12 and 13 years, 18% 14 years, 29.3% 15 years, 34.7% 16 years, 12% 17
years, and 3.4% 18 years). Some who committed a crime or were brought to
court for problem behaviours when they were 17 were adjudicated after their
18th birthdays. In sum, 87.4% of the girls were 14, 15, 16 and 17 years old.
These girls originated from broken homes (only 18% lived with their two
biological parents) and most of their parents were born in Quebec (91%). Of
the adjudicated girls, 51% were gang members, but very few were members of a
structured gang. In sum, this sample included girls and boys with important risk
factors for delinquency and problem behaviour.
The sample of deviant behaviours is composed of 61 questions covering the
10 types of deviant behaviour displayed in Figure 1. Sixteen behaviours had to
be removed from the model fitting procedure for the following reasons. Seven
behaviours were discarded because their bad kurtosis produced very low
loadings on the obtained factors even if these behaviours were loading the
hypothesis factor (sexual aggression, glue sniffing, homosexual and hetero-
sexual prostitution, homosexual sex, hitting parents, fire setting). In addition,
CBMH 13.2_2nd 3/10/03 5:54 pm Page 93

Development test of the general deviance syndrome 93

nine items were eliminated because of their very low loading even if it was on
the hypothesis factor (bicycle theft, theft at school, vandalism at school, false
alarm, obscene phone call, gambling, using pornographic material, accusing
others of starting a fight, and hitting siblings). In consequence, 45 behaviours
were retained for model testing. For each of these behaviours, the frequency
during the last year and the age of onset were asked at each interview. The
average ages of onset are reported in Table 2. All the data are right censored by
this coding procedure for missing data (the age of the oldest subject plus one)
and by the age at last interview. In addition, it has been empirically established
that self-reported deviance data from adolescents involved in the criminal
justice system are reliable and valid (Hindelang et al., 1981; Cohen, 1986).
The following four objectives are pursued in this paper. The first objective
was the identification of the general deviance latent construct with the
procedure used by 15 of the 21 previous studies. In consequence, we used nine
scales to perform three-level confirmatory factor analysis (CFA) using the last
year’s frequency data. The second objective was the developmental test of the
general deviance syndrome using the age of onset. The third objective was the
simultaneous testing of the unidimensional and multidimensional nature of
deviance. In this case, we performed a four-level CFA using the 45 behaviours,
nine types and four forms of deviant behaviour, and the general deviance
construct. The fourth objective was the independent test of the general
deviance syndrome for girls and boys. In this case, we included a group
comparison procedure in CFA. To attain these objectives, we employed CFA
calculated with the EQS program for the Macintosh computer (Bentler, 1995).

Results

Using the frequency of the 45 deviant behaviours grouped into 12 types, the
model could not be fitted satisfactory (these data are available from the first
author). This conclusion was reached because the confirmatory factor analysis
did not meet the usual fit criterion for a model, that is when the chi-square
statistic is less than two times the degrees of freedom and the comparative fit
index (CFI) and the normed fit index (NFI) are greater than 0.90 (Bentler,
1995). In addition, a number of loadings (the standardized contribution to a
factor by each behaviour) on the 12 types of deviant behaviour were very low
for some behaviours. This situation may be explained by the following reasons:
the number of behaviours representing a type was often low, sometimes
composed of only two behaviours; the number of types, 12, largely exceeded
existing tests of such a model in the literature by a factor of more than two; the
number of subjects reporting a behaviour was very small for some behaviours
even in a sample of adjudicated adolescents (for example rape); some correla-
tions between behaviours were very small. Because of these difficulties, we
performed item analysis of the scales in Figure 1. These analyses indicated that
CBMH 13.2_2nd 3/10/03 5:54 pm Page 94

94 Le Blanc, Bouthillier

the 12 types of deviant behaviour had to be reduced to nine; some scales could
be grouped in one: overt and covert family rebellion, intimidation and physical
violence, and minor and major theft.
The first step of our analytical strategy called for the test of the existence of
the general deviance latent construct, a replication of the studies reported in
Table 1. Figure 2 displays these results. The hierarchical CFA with the nine
scales and the four forms of deviance fitted well the frequency data (NFI 0.90,
CFI 0.91). This time the general deviance syndrome was replicated with an
adjudicated sample outside the United States.

Figure 2. Frequency, Hierarchical CFA of general deviance


CBMH 13.2_2nd 3/10/03 5:54 pm Page 95

Development test of the general deviance syndrome 95

The second step of our analytical strategy called for a developmental test of
the general deviance syndrome. Figure 3 reports our replication of the general
deviance latent construct with the age of onset on the nine scales and four
forms of deviant behaviour. The test is clearly positive (NFI 0.92, CFI 0.95).
These indexes are even higher than for the model with the frequency measure
above. In consequence, the general deviance syndrome can be identified using
a developmental indicator of deviance as well as when employing a usual
frequency indicator.

Figure 3. Onset, Hierarchical CFA of general deviance


CBMH 13.2_2nd 3/10/03 5:54 pm Page 96

96 Le Blanc, Bouthillier

The third step of our analytical strategy consisted in the simultaneous test
of the unidimensional and multidimensional nature of the general deviance
syndrome. This test involves a four-level hierarchical CFA, 45 behaviours,
nine types and four forms of deviant behaviour, and the general latent
construct. Existing studies limit themselves to two levels: at the first level
either specific behaviours or most often types, and at the second level the
general latent construct. They were adequate tests of the unidimensional
nature of the general deviance syndrome. However, they did not perform a test
of the multidimensional nature of deviance. Our four-level CFA offers the
usual unidimensional test with the types of deviant behaviours and the general
latent construct and the multidimensional test with the 45 specific behaviours
grouped under the nine types and four forms of deviant behaviour. When we
conducted this four-level hierarchical CFA with the frequency (NFI 0.80, CFI
0.86) and the onset (NFI 0.84, CFI 0.85) data, the fit indexes were less satis-
factory (these results are not reported in a figure because it would be composed
of the content of Figures 2 and 4 for example). These moderate results ensue
from the multidimensional component of the model, the 45 behaviours
composing the nine forms of deviance, rather than the hierarchical
component, the nine types of deviance combined in four forms and in the
general construct. Figures 4 and 5 illustrate this phenomenon since the fit
indexes for all the models tested are in the 80s and even lower for girls (for the
frequency data: total sample: NFI 0.80, CFI 0.86; boys: NFI 0.77, CFI 0.85;
girls: NFI 0.69, CFI 0.74; for the onset data: total sample: NFI 0.79, CFI 0.85;
boys: NFI 0.76, CFI 0.84; girls: NFI 0.60, CFI 0.78).
The possible reasons for these results are routinely discussed in the liter-
ature (Long, 1983; Bentler, 1995). First, our sample, 656 subjects, may be
insufficiently large considering these four hierarchical levels. In addition,
missing data were increasing rapidly with the list-wise deletion when using the
45 behaviours. Second, our data may not be perfectly adequate because there
were sometimes only a few behaviours composing a scale, as explained earlier.
Third, the number of subjects reporting some behaviours, as we have observed,
was small even in an adjudicated sample. Fourth, the model was possibly
misspecified even if we used item analysis to group behaviours in types of
deviance and even if we tested various combinations of types of deviance to
construct the forms of deviance. Fifth, the starting values for iterations may
have been inappropriate since completely satisfactory solution of this problem
did not exist. Sixth, and finally, we observed that some correlations between
behaviours were two small in comparison with others that were much higher
without been excessively high. Notwithstanding these limits of our test of a
very comprehensive four-level model of general deviance, we were confident
that our results were a powerful and clear indication that our theoretical
general deviance model was sound. It is reasonable to expect that this model
could be successfully fitted with a greater number of behaviours and a larger
sample.
CBMH 13.2_2nd 3/10/03 5:54 pm Page 97

Development test of the general deviance syndrome 97

F1 Sexual
activity and
alcohol

F2
Substance
use

F3
Authority conflict
in school

F4
Authority conflict
at home

F5
Fraud

F6
Theft

F7
Motor
vehicle use

F8
Vandalism

F9
Violence

Figure 4. Frequency, Confirmatory factor analysis of the 45 deviant behaviours


CBMH 13.2_2nd 3/10/03 5:54 pm Page 98

98 Le Blanc, Bouthillier

F1 Sexual
activity and
alcohol

F2
Substance
use

F3
Authority conflict
in school

F4
Authority conflict
at home

F5
Fraud

F6
Theft

F7
Motor
vehicle use

F8
Vandalism

F9
Violence

Figure 5. Onset, Confirmatory factor analysis of the 45 deviant behaviours


CBMH 13.2_2nd 3/10/03 5:54 pm Page 99

Development test of the general deviance syndrome 99

The last step of our analytical strategy was to replicate the presence of the
general deviance latent structure for girls and boys. This was done indepen-
dently for girls and boys as reported in parentheses in Figures 2 to 5 and
through CFA with a group comparison on gender. The results were satisfactory
for the frequency data (boys: NFI 0.90, CFI 0.92; girls: NFI 0.92, CFI 0.95).
The group comparison on gender was also satisfactory for the frequency data
and this test was never reported in the literature (equal loadings: NFI 0.88, CFI
0.90). The results were comparable for the onset data (boys: NFI 0.92, CFI
0.95; girls: NFI 0.77, CFI 0.91) and for the group comparison (equal loadings:
NFI 0.88, CFI 0.95). However, the fit indexes were less satisfactory when 45
behaviours, nine types, four forms of deviant behaviour, and general deviance
were used (boys: NFI 0.77, CFI 0.85; girls: NFI 0.58, CFI 0.74; group
comparison with equal loadings: NFI 0.88, CFI 0.95).
In sum, in some tests the results for the adjudicated girls were less satis-
factory than for the boys. In the case of girls, EQS indicated a parameter
condition code because a factor, property crimes, was constrained at a lower
bound (the factor covert had a loading of 1). In this sort of case, Bentler and
Chou (1987) indicated that we might not worry if the results make sense and
the Heywood cases (over-factoring) were isolated occurrences. This situation
may have happened because the sample size was too small for adequate reliance
on large sample theory, which was the case with girls (there were 150 girls and
some 171 parameters in the model). In this context, Bentler and Chou (1987)
recommended the acceptance of the model. Another explanation of our diffi-
culties in fitting the model for girls was what Rindskopf (1987) calls
over-factoring: we may have used too many types of deviance. Since the partic-
ipation in delinquent behaviour is low for girls, it is possible, for example, that
the onset of theft may explain a large proportion of variance of the latent
factors overt and covert deviance.
An additional explanation of the fact that the results were less satisfactory
for girls may rest on the fact that the limits of the girls’ data were more serious
than the limits of the boys’ data, particularly the sample size and the left-hand
distribution of the subjects on the behaviours. There are some major differ-
ences between girls and boys in their respective level of participation,
frequency, seriousness and variety (Lanctôt and Le Blanc, 2002); in conse-
quence, some authors think that there should be specific measures of deviance
for each gender. We observed that the constraints for the models indicated
that several loadings were significantly different for boys and girls on the
univariate tests (motor vehicle use, theft, covert behaviour and authority
conflict). Accordingly, we tried to fit models in which most parameters were
constrained to be the same for girls and boys and only some were allowed to
differ instead of testing the model with equal loadings for both genders as we
did. This strategy of model fitting was retained because the former strategy led
to numerous tests and there were no clear empirical or theoretical reasons to
choose which parameters should differ. In sum, the fact that the level of partic-
CBMH 13.2_2nd 3/10/03 5:54 pm Page 100

100 Le Blanc, Bouthillier

ipation, frequency, onset and so on is lower for girls than boys does not mean
that the structure of the general deviance syndrome is specific for each gender.
Our results can be interpreted as an indication that the structure of the general
deviance syndrome was invariant across genders, but there was still a need to
use some parameters that were different for girls and boys in the structural
model. The results were a clear indication that the same scales can measure
deviance, but that the weight of the particular variables will differ for girls and
boys. Although the comparison for the frequency model indicated some differ-
ences between boys and girls, the onset model applied reasonably well in both
groups, even though a difference was noted for one parameter (theft). This is a
stronger argument for the absence of a gender gap in the structure of the
general deviance syndrome. In addition, these results confirmed that there was
a measurement invariance of the general deviance syndrome for girls and boys.
These results can be interpreted as an indication that the structure of the
general deviance syndrome for onset was invariant across genders, but that
there was a need to use some parameters that were different for girls and boys
in the structural model of the frequency data. The results were an indication
that the same scales measured the construct of general deviance. But since the
weight of particular variables differed for girls and boys in the frequency
models, this may indicate that the models measured a somewhat different
construct for one of the groups.

Discussion and conclusion

In the present study, four objectives were pursued. The first objective was the
cross-cultural replication of the latent construct of general deviance. The
second objective consisted in a developmental test of the general deviance
syndrome. The third objective referred to the simultaneous testing of the
unidimensional and the multidimensional nature of general deviance. Finally,
the fourth objective was the comparison of the structure of the general
deviance syndrome for boys and girls.
Before discussing our results, some limits of our data and results should be
highlighted. We employed retrospective self-reported data of current frequency
and age of onset of deviant behaviours. These data are subject to the well-
known retrospective biases. However, we should mention that it is interesting
to note that our interviewers reported that they were always surprised how easy
it was for the subjects to remember the age of the first time they were involved
in a particular deviant behaviour and described the circumstances of that
behaviour (nature, place, accomplices and so on) (see Le Blanc and Fréchette,
1989). The overall sample size (656) was adequate for our hierarchical confir-
matory factor analyses, but 150 was not totally sufficient for the girls. In an
adjudicated sample, the participation and the frequency of the perpetration of
deviant behaviour are very much higher than in a representative sample for all
types of deviant behaviours. In addition, a reasonable rate of participation and
CBMH 13.2_2nd 3/10/03 5:54 pm Page 101

Development test of the general deviance syndrome 101

frequency is also available for more serious deviant and delinquent acts and, in
consequence, a larger array of deviant behaviours can be considered. As for the
confirmatory factor analyses, we observed that the results were less satisfactory
when we used the behaviours instead of the scales for girls. In the first case, we
have to acknowledge that our sample could have been too small considering
the number of variables (45) and levels (3) and that for some scales the behav-
iours composing them were only a few. In addition, the models were possibly
misspecified even if we used factor analysis and item analysis to group behav-
iours in types of deviance and even if we tested various combinations of types
of deviance to compose the forms of deviance. In the case of the tests with
girls, there were some weaknesses even if we can retain confidence in the fits.
The sample was small in relation to the number of behaviours and levels in the
confirmatory factor analyses; this was accentuated by missing data. The
number of subjects reporting some behaviours was small and some distributions
were special for girls. Notwithstanding all the above limits of our test of a very
comprehensive four-level model of general deviance, we are confident that our
results suggested powerful and clear indications that our general deviance
theoretical structure was sound.
Twenty-one previous studies confirmed the existence of the general
deviance syndrome for samples of different natures. Only two studies arrived at
the same conclusion outside the United States, in England (Emler and
Reicher, 1995) and in New Zealand (Bartush et al., 1998). Our positive results
with a French-speaking sample seem to indicate that we can expect similar
results in other occidental countries. We also know from cross-cultural
comparisons that the epidemiology of deviance was very similar when compa-
rable samples were used (Junger-Tas et al., 1994). In addition, two studies
confirmed the general deviance latent construct with adjudicated samples of
boys. Our result with boys and girls is a supplementary corroboration of this
construct. In addition, we can have strong confidence in the existence of the
general deviance syndrome because the studies were conducted during the
1970s, 1980s and 1990s and one study made a direct comparison of adjudicated
samples recruited during the 1970s and 1990s with the same set of behaviours.
Furthermore, it has been shown repeatedly that the results are independent of
the size of the representative sample and its setting – metropolitan, urban,
rural, regions of the US and so on. Notwithstanding all these corroborations,
we can conjecture that the structure of deviance is universal.
All previous studies performed a cross-sectional test of the general deviance
hypothesis with data on frequency or participation in deviance. Our test with
data on the timing of onset confirms that it is also a developmental
phenomenon. We knew from previous studies that a latent construct could be
established for different age groups during childhood, adolescence and
adulthood. However, we did not know whether the general deviance syndrome
manifested itself as a sequence of behaviour. It is definitively the case. The
general deviance syndrome appears as the co-occurrence of various deviant
CBMH 13.2_2nd 3/10/03 5:54 pm Page 102

102 Le Blanc, Bouthillier

behaviours at a specific point during the life course, but also as a sequence of
initiations of different deviant behaviours. This is additional confirmation of
the usefulness of a developmental paradigm for the understanding of deviance
in general, not only for offending for which the demonstration was solid (Le
Blanc and Loeber, 1998). The strength of our developmental test of the
sequential initiation of deviant behaviour lay in the range of behaviours, 45,
and types, nine, and the complexity of the structure of the construct, four
levels. However, this test does not shed light on the controversy between the
heterotypic and the homeotypic continuity of general deviance. As we
indicated, one study (Bartush et al., 1998) favours the first possibility and
another study (McGee and Newcomb, 1992) the second point of view. To
enlighten that debate, we need a simultaneous test of the two possibilities.
This test should involve the comparison, on the same data set, of a model with
similar behaviours at different ages and a model with different behaviours at
different ages during the life course.
Theoretical criminology has always been quartered between the defenders
of general and specific theories, between explanations that could apply to all
forms (Gottfredson and Hirschi, 1990, for example) or to specific forms of
deviance (for example Cullen, 1985). Accordingly, empiricists alternated
between an omnibus scale and multiple scales to represent the range of deviant
behaviours. There were numerous proofs of the two positions that we reviewed.
Empiricists recurrently showed that many behaviours could be grouped into a
small number of types of deviance and that a number of types of deviant
behaviour formed a latent construct. In addition, they demonstrated
repeatedly that general theory, such as control and learning explanations, were
valid, but that some variables may be more or less significant for various types
of deviance. Our simultaneous tests of the unidimensional and multidimen-
sional nature of general deviance with 45 behaviours and nine types subsumed
under four forms of deviance and the global construct indicated that general
deviance was both at the same time. In consequence, theories must be able to
account for the unidimensional and multidimensional nature of general
deviance. For example, differential association theory should be stated in
general terms: that the affiliation to deviant peers and the learning of defini-
tions favourable to deviance increase general deviance. In addition, it should
state that these affiliations and definitions must be specific for particular types
of deviance, and that they should be directed at delinquent peers for delin-
quency, drug user peers for drug use, school dropout peers for school dropout
and so on.
Fifteen of the 21 studies reviewed used mixed-gender samples. However,
only three of them tested the general deviance model by gender and it was on
the same group of representative samples. Our tests of the general deviance
syndrome showed that we could assume that it is composed of the same types
and forms of deviant behaviours for females and males. The measurement of
general deviance can be accomplished in the same way for boys and girls.
CBMH 13.2_2nd 3/10/03 5:54 pm Page 103

Development test of the general deviance syndrome 103

However, the weights of the variables will change for genders. The gender gap
in deviant behaviour, documented by the empirical literature (see Lanctôt and
Le Blanc, 2002), is only a question of the level of participation, frequency,
variety and so on; it is not a question of the structure of the general deviance
syndrome.
Our study improved over existing tests of the general deviance syndrome by
using a wider range of deviant behaviours, 45, more types of deviance, nine,
and more levels in our hierarchical confirmatory factor analyses, three or four
instead of two. However, we could expect that it would be very difficult to
conduct such test with a representative sample for the following reasons. The
participation in some deviant behaviours, particularly the most serious ones,
will be too low in such a sample. In consequence, only purposive samples, such
as adjudicated ones, can be used for the replication of our test of our compre-
hensive general deviance model. With our sample size, we barely could test our
model reduced to 45 behaviours, nine types and four forms of deviance. We are
confident that with a much larger sample we could test our theoretical model
that was composed of 61 behaviours, 12 types and four categories of deviance,
and the latent construct.

Acknowledgement

This work was supporte by le Conseil de recherches en sciences humaines du


Canada, le Conseil québécois de la recherché social et le Fond pour la
formation des chercheurs et l’action concertée du Québec.

References
Achenbach TM (1985) Assessment and Taxonomy of Child and Adolescent Psychopathology.
Beverly Hills, CA: Sage.
Ary DV, Duncan TE, Biglan A, Metzler CW, Noell JW, Smolkowski K (1999) Development of
adolescent problem behavior. Journal of Abnormal Child Psychology 27: 141–150.
Bartush DRJ, Lynam DR , Moffitt TE, Silva PA (1997) Is age important? Testing a general
versus a developmental theory of antisocial behavior. Criminology 35, 13–48.
Benda BB, Corwyn RF (1998) Adolescent deviant behavior: multiple contingency table
analyses of overlap between behaviors. Journal of Social Science Research 24, 29–59.
Bentler PM (1995) EQS Structural Equation Program Manual. Encico Multivariate Software.
Bentler PM, Chou C-P (1987) Practical issues in structural modelling. Sociological Methods and
Research 16, 78–117.
Brownfield D, Sorenson AM (1987) A latent structure analysis of delinquency. Journal of
Quantitative Criminology 3(2), 103–124.
Cappaldi DM, Patterson GR (1989) Psychometric Properties of Fourteen Latent Constructs for the
Oregon Youth Study. New York: Springer-Verlag.
Cheong YF, Raudenbush SW (2000) Measurement and structural models of children’s problem
behaviors. Psychological Methods 5, 477–495.
Cohen J (1986) Research on criminal career: individual frequency rates and offence seriousness
In Blumstein A, Cohen J, Roth JA, Visher CA. Criminal Careers and Career Criminals
Washington, DC: National Academy Press.
CBMH 13.2_2nd 3/10/03 5:54 pm Page 104

104 Le Blanc, Bouthillier

Corwyn RF, Benda BB (1999) Multiple contingency table analyses of the deviance syndrome:
how much overlap there is? Journal of Child and Adolescent Substance Abuse 9, 39–56.
Cullen FT (1985) Rethinking Crime and Deviance Theory. Totowa: Rowman & Allanheld.
Dembo R, Williams L, Wothke W, Schmeidler J, Getreu A, Berry E, Wish ED (1992) The
generality of deviance: replication of a structural model among high-risk youths. Journal of
Research in Crime and Delinquency 29(2), 200–216.
Donovan JE, Jessor R (1985) Structure of problem behavior in adolescence and young
adulthood. Journal of Consulting and Clinical Psychology 53(6), 890–904.
Donovan JE, Jessor R, Costa FM (1988) Structure of problem behavior in adolescence: a repli-
cation. Journal of Consulting and Clinical Psychology 56(5), 762–765.
Elliott DS, Huizinga D, Ageton SS (1985) Explaining Delinquency and Drug Use. Beverly Hills:
Sage Publications.
Emler N, Reicher S (1995) Adolescence and Delinquency. Oxford: Blackwell.
Farrell AD, Kung EM, White KS, Valois RF (2000) The structure of self-reported aggression
drug use and delinquent behaviour during early adolescence. Journal of Clinical Child
Psychology 29, 282–292.
Frick PJ, Kamphaus RW, Lahey BB, Loeber R, Christ MAG, Hart EL, Tannenbaum LE (1991)
Academic underachievement and the disruptive behavior disorders. Journal of Consulting and
Clinical Psychology 59, 289–294.
Frick PJ, Lahey BB, Loeber R, Tannenbaum L, Van Horn Y, Christ MAG, Hart EL, Hanson K
(1993) Oppositional defiant disorder and conduct disorder: a meta-analytic review of factor
analyses and cross-validation in a clinic sample. Clinical Psychology Review 13, 319–340.
Gilmore MR, Hawkins JD, Catalano RF, Day LE, Moore M, Abbott R (1991) Structure of
problem behaviors in preadolescence. Journal of Consulting and Clinical Psychology 59(4),
499–506.
Gottfredson MA, Hirschi T (1990) A General Theory of Crime Stanford, CA: Stanford
University Press.
Hindelang MJ, Hirschi T, Weis JG (1981) Measuring Delinquency. Beverly Hills, CA: Sage.
Jessor R, Jessor SL (1977) Problem Behavior and Psychosocial Development New York: Academic
Press.
Johnston LD, O’Malley PM, Eveland LK (1978) Drugs and delinquency: a search for causal
connections. In Kandel DB, ed. Longitudinal Research on Drug Use: Empirical Findings and
Methodological Issues. Washington, DC: Hemisphere.
Junger-Tas J, Terlouw G-J, Klein MW (1994) Delinquent Behaviour among Young People in the
Western World. Amsterdam: Kugler.
Kulik JA, Stein KB, Sarbin TR (1968) Dimensions and patterns of adolescent antisocial
behaviour. Journal of Consulting and Clinical Psychology 32, 375–382.
Lanctôt N, Le Blanc M (2002) Explaining adolescent females’ involvement in general deviance:
towards an integration of theoretical perspectives. Crime and Justice 26: 113–202.
Le Blanc M (2002) The cycle of antisocial behaviour: escalation and de-escalation. International
Journal of Comparative and Applied Criminal Justice 26: 53–84.
Le Blanc M, Girard S (1997) The generality of deviance: replication over two decades with a
Canadian sample of adjudicated. Canadian Journal of Criminology 39, 171–183.
Le Blanc M, Janosz M (1998) The development of problem behavior: course and processes.
Seattle Life History Research Society, 6–9 May.
Le Blanc M, Loeber R (1998) Developmental criminology upgraded. In Tonry M, Morris N, eds.
Crime and Justice Handbook. Chicago: University of Chicago Press, Vol. 23, 115–198.
Le Blanc M, Fréchette M (1989) Male Criminal Activity from Childhood Through Youth: Multilevel
and Developmental Perspectives. New York: Springer-Verlag.
Loeber R, Le Blanc M (1990) Toward a developmental criminology. In Tonry M, Morris N, eds.
Crime and Justice. Chicago, IL: University of Chicago Press, Vol. 12, pp 375–473.
CBMH 13.2_2nd 3/10/03 5:54 pm Page 105

Development test of the general deviance syndrome 105

Loeber R, Farrington DP, Stouthamer-Loeber M, Van Kammen WB (1998) Antisocial Behaviour


and Mental Health Problems: Risk Factors in Childhood and Adolescence. London: Cambridge
University Press.
Loeber R, Schmaling K (1985) Empirical evidence for overt and covert patterns of antisocial
conduct problems. Journal of Abnormal Child Psychology 13, 337–352.
Long JS (1983) Confirmatory Factor Analysis. Newbury Park, CA: Sage.
McGee L, Newcomb MD (1992) General deviance syndrome: expanded hierarchical evalua-
tions at four ages from early adolescence to adulthood. Journal of Consulting and Clinical
Psychology 60(5), 766–776.
Nye FI, Short JF (1957) Scaling delinquent behaviour. American Sociological Review 22,
326–331.
Osgood DW, Johnston LD, O’Malley PM, Bachman JG (1988) The generality of deviance in
late adolescence and early adulthood. American Sociological Review 53, 81–93.
Parker RN, McDowell D (1986) Constructing an index of officially recorded crime: the use of
confirmatory factor analysis. Journal of Quantitative Criminology 2(3), 237–250.
Quay HC, Blumen L (1963) Dimensions of delinquent behavior. Journal of Social Psychology 61,
273–278.
Rindskopf D (1984) Structural equation models, empirical identification, Heywood cases and
related problems. Sociological Methods and Research 13, 109–119.
Scott JF (1959) The dimensions of delinquent behavior. American Sociological Review 2,
240–243.
Ullmann JB, Newcomb MD (1999) The transition from adolescent to adult: a time of change in
general and specific deviance. Criminal Behaviour and Mental Health 9,: 74–90.

Address correspondence to: Marc Le Blanc PhD (Criminology) SRC École de


psychoéducation and École de criminologie Université de Montréal CP 6128
succ Centre Ville Montréal Québec Canada H2C 3J7 Email: marcleblanc@
umontrealca

Das könnte Ihnen auch gefallen